Abstract: Color histogram image retrieval, which is based on the color character, is a very effective image retrieval method. In this paper, image similarity is defined as the intersectant area between two gray level probability histograms. It ranges from 0 to 1. This concept is more specific to describe the image similarity. Some groups of remotely sensed data's image similarities are computed to test whether this method is valid or not. The result shows that this method can discriminate the images from each other easily, which is more effective than the information entropy method to compute image similarity. As a whole, it is a feasible method.

Key words: similarity, histogram, area, information entropy, image retrieval